# AI Assistants Are Flattening the Diversity of Human Writing

> Large language models polish our text, but new research shows they erode the rich diversity and individual cues hidden in human writing.
- Title: AI Assistants Are Flattening the Diversity of Human Writing
- Summary: Large language models polish our text, but new research shows they erode the rich diversity and individual cues hidden in human writing. Using AI to edit…
- Keywords: linguistics, psychology, writing, llms, technology, startups, Assistants, Flattening, Diversity, Human, Nature
- Source: Nature — https://www.nature.com/articles/s41562-026-02550-0
- Read time: 1 min
- Topics: ai, linguistics, psychology, writing, llms, technology, startups
## Language carries subtle human signals
Human writing does far more than convey basic facts. It encodes rich data about identity, mental state, and social context, providing critical signals for healthcare and social sciences.

> Language is far more than a communication tool; it encodes a wealth of information about a person’s identity, psychological state and social context...
## AI editing erodes linguistic diversity
A study across seven datasets and 880,000 texts shows that adopting large language models as writing tools correlates with a widespread drop in writing variety.

> ...the widespread adoption of large language models (LLMs) as writing assistants is linked to declines in linguistic diversity...
## Writing styles are becoming homogenized
While AI tools successfully preserve core ideas when polishing text, they significantly flatten complexity. Writing-complexity variance dropped by 21% to 50% across tested models.

> ...LLMs also homogenize writing styles, reducing writing-complexity variance by a statistically significant 21–50% across datasets and models...
## Conformity replaces unique individuality
LLM rewriting systematically highlights dominant traits while smoothing over unique or minority characteristics, favoring standardized conformity over individual voice.

> ...amplify patterns associated with dominant characteristics while suppressing others, emphasizing conformity over individuality.
## Standardization threatens real-world evaluations
As individual voice vanishes into AI polish, key diagnostic processes, personalized services, hiring assessments, and efforts to preserve cultural nuances face serious challenges.
## Key takeaway

Using AI to edit writing smooths out unique stylistic signals, which could compromise psychological assessments, hiring evaluations, and cultural preservation.